Many aspects of the social world bear on important health outcomes, but for most, the exact nature of their impact is still unclear. Although much has been learned through statistical approaches, the literature falls short of a causal understanding of how macro and micro processes interrelate in affecting health. The proposed research uses a spatially explicit agent-based modeling approach, informed by insights from sociology, geography, and economics. Dynamic social networks are included endogenously, an especially innovative element. Through the construction of a model in one setting, a set of modeling tools will be developed that will be applicable across a wide range of settings. These new tools will be made available through a public website. Once the agent-based model is constructed, it will be used to study interconnections among the social, spatial, and biophysical dimensions of the local context, taking into account migration, residential choice, land use, and household wealth, and feedbacks therein. In the process will come a better understanding of the consequences of oversimplification in standard statistical models of community effects. The model will also be used to conduct experiments about the effects of a sudden shift in infant mortality and shifts in economic conditions. The model will be field tested twice, the first time to validate key model assumptions and the second time to explore unexpected results. Indeed, an advantage of the agent-based approach is the possibility of "emergence" at the system or community level (i.e., the integration of macro and micro processes producing new structures not anticipated based on a simple aggregation of individual and household behaviors). Finally, borrowing and extending tools from the fields of meteorology and control theory, new methods for testing model sensitivity that are more computationally efficient than those now in use will be developed, applied, and made broadly available. The proposed project will develop tools to study social processes involving individuals, households, social networks, and communities in relation to health. The application of these tools will help us better understand and interpret the research literature connecting community factors with health outcomes and provide a complement to the standard statistical approaches. [unreadable] [unreadable] [unreadable]